392 METALLURGY OF CAST IRON.
always the same for the same grade of iron, tinder the conditions of these tests; so that, however "hot" iron may have been poured, it will always have a certain temperature when it begins to expand. But it is, of course, clear that expansion will take place sooner in a "dull "-poured bar than in a "hot" one; and again, a light body will expand more quickly than a heavy one, as I have proved by my tests.
The length of the period of expansion varies with the size of the casting. The more massive the casting, the longer the period of expansion. In the bars shown in Figs. 74 and 75, the expansion lasted from one-half to one minute in the smallest bars, and, in the largest bars, from three to five minutes. The relation between the shrinkage and the expansion of solidification may now be indicated. The author's view is that the apparent shrinkage of liquid metal so familiar to heavy founders is not clue chiefly to a change in the specific gravity of the liquid metal as it passes to a solid state, but largely to the effect of the expansion of the solidifying parts of the casting. That is to say, an outer shell of the casting being first formed, its expansion at the moment of solidification necessarily enlarges the interior space to be occupied by liquid metal; and either additional liquid metal must be applied or else cavities and shrink-holes will be found in the interior of medium and heavy castings, by reason of the progressive accretion of the solidifying metal upon the parts already solidified. Such cavities would, on this hypothesis, be likely to be most abundant in the portions which solidify last; and that this is in fact the case, is often proved by practice. Cavities are very liable to occur in the interior ofasonable to say, isU Citut